Strategists from both parties agree that Sen. Rick Santorum (R) faces an extremely difficult race against his Democratic challenger, state Treasurer Bob Casey Jr. (D), in next year’s Pennsylvania Senate race.

For the moment, let’s forget about the impact of the candidates’ money, President Bush, Santorum’s new book and a host of other factors. Let’s look just at Santorum’s poll numbers, and at how he compares with other incumbent Senators who went down to defeat during the past three Congressional elections.[IMGCAP(1)]

Is Santorum better positioned at this point than Sens. Rod Grams (R-Minn.) and Chuck Robb (D-Va.) were a year and a half before their re-election races? Is he stronger in the polls than Spencer Abraham (R-Mich.), Tom Daschle (D-S.D.), Slade Gorton (R-Wash.), Tim Hutchinson (R-Ark.) and Max Cleland (D-Ga.) were the summer before their re-election contests?

The first number everyone looks at — not wisely, I might add — is the ballot test.

Three different polls, one conducted by Franklin and Marshall College and two by Quinnipiac University, show Santorum trailing Casey, by 7, 11 and 14 points.

Very few incumbents trail their opponents at this point in the cycle, so Santorum’s initial standing in the ballot test is horrendous. But a comeback wouldn’t be unprecedented. In the summer of 1983, a year and a half before the general election, incumbent Sen. Jesse Helms (R-N.C.) was trailing Gov. Jim Hunt (D-N.C.) by almost 20 percentage points in the Senate race. Yet Helms came back to win by 4 points.

Santorum and his allies argue that he trails Casey because the Democrat’s father was a beloved figure in the state and the Casey name is a huge asset. They believe the polls will change when voters meet Casey Jr. and compare him with Santorum.

Santorum’s argument strikes me as reasonable, but not entirely convincing. Yes, Casey has assets that many challengers don’t, but incumbency is also a powerful weapon, which is why most of the Senators who lost re-election over the past five years held leads in the polls 18 months before voters dumped them.

Grams, Cleland and Daschle led in all public polls conducted the year before each lost. Abraham and Gorton held leads in most, though not all, surveys conducted the year before they faced voters.

In March 2001 then-Attorney General Mary Pryor (D) held a 2-point lead in a Democratic poll over incumbent Hutchinson (R) in the 2002 Arkansas Senate race. Other surveys, however, showed the Republican with an edge within the margin of error. He ultimately lost.

Perhaps the best analogy to this cycle’s Santorum race is the 2000 contest between then-Sen. Chuck Robb (D-Va.) and former Virginia Gov. George Allen (R).

Polling by Mason-Dixon throughout 1999 showed Allen — a challenger who also had considerable assets when he began his Senate campaign — with a double-digit lead over the incumbent. In the June 1999 survey, Allen led Robb 49 percent to 38 percent, and in September, Allen was ahead 50 percent to 38 percent.

Given Allen’s 5-point win over Robb in the 2000 elections, the similarity between the early Virginia ballot tests and the current Pennsylvania Senate polls suggest that Santorum will need a Helms-like rally to hold his seat.

Santorum has noted that while he trails in the ballot test, 45 percent of those polled say he deserves re-election, compared to only 38 percent who say he doesn’t.

“How many members have a 45 percent re-elect,” he asked rhetorically to a Roll Call reporter. Re-elects aren’t always asked or made public, but the answer, from a search of the Hotline’s current poll database, is five: John Kyl (R-Ariz.), Dianne Feinstein (D-Calif.), Joe Lieberman (D-Conn.), Edward Kennedy (D-Mass.) and Hillary Rodham Clinton (D-N.Y.). All five are expected to win re-election, several of them quite easily.

Re-elects, however, aren’t the only measure of an incumbent’s strength. Santorum’s job approval in the July Quinnipiac poll was 51 percent approve, 35 percent disapprove — good numbers that probably help explain his re-elect scores.

The Pennsylvania Republican’s 51 percent job approval is better than Abraham’s job rating in August 1999, when 39 percent of those polled gave the Michigan Republican a positive rating and 33 percent rated his performance as negative. But 57 percent of respondents gave Daschle a good or excellent job rating in August 2003 — an even stronger showing than Santorum’s. And Daschle, as everyone knows, eventually lost.

But Santorum’s personal ratings are considerably less impressive. His 37 percent favorable, 27 percent unfavorable ratings in last month’s survey aren’t much different than Robb’s 44 percent favorable, 27 percent unfavorable in September 1999, or Grams’ 38 percent favorable, 33 percent unfavorable in July of that year.

The Pennsylvania Republican’s numbers are not as good as Abraham’s were in August 1999 (41 percent favorable, 25 percent unfavorable) or Cleland’s were in June 2001 (52 percent favorable, 20 percent unfavorable).

Last year, voters in South Dakota liked both Daschle and John Thune (R-S.D.). Those red state voters ended up voting for the Republican. Santorum isn’t as well positioned as Daschle, whose job rating with 56 percent good or excellent and name ID was 54 percent favorable the year before he was defeated.

Santorum’s re-elect offers the Republican hope in his bid for another term. But his other numbers, when viewed in light of previous incumbent losses, confirm his serious problems. And like Daschle, Grams, Abraham, Cleland and Gorton — and unlike Helms — Santorum’s party label doesn’t match the color of his state.

Stuart Rothenberg is editor of the Rothenberg Political Report